改进YOLOv5的多样态木蜂孔洞检测算法OA

In view of the problem of identifying the types and shapes of cross-sectional holes in building compo-nents,this study proposes an improved YOLOv5(You Only Look Once version 5)multi-state wood bee hole detec-tion algorithm.The coordinate attention mechanism is integrated in the model backbone network to enhance the fea-ture extraction ability,and the weighted BiFPN(Bi-directional Feature Pyramid Network)is combined to optimize the multi-scale feature integration.The detection head adopts a decoupled structure,separating classification and re-gression tasks,and introduces the loss function EIoU(Efficient Intersection over Union)to enhance the accuracy of target positioning and the convergence efficiency of the model.The experimental results show that the mAP(mean Average Precision)of the improved YOLOv5 model in the wood bee hole detection task is 99.46%.While enhancing the detection accuracy,the proposed method maintains the lightweight of the model,achieving high-precision and high-efficiency detection of wooden honeycomb holes,which is of great value for the assessment and repair of damage grades in wooden buildings.